Research Cycle

Use the parts needed for the current scientific decision inside the existing task. Routine progress checks, resuming an unchanged protocol, ordinary implementation, and unchanged status reports do not trigger this skill. Do not reread it or open a research map just to satisfy a heartbeat.

Reuse the current question, protocol, sources, and analysis already available. Fill decision-relevant gaps; do not rewrite complete material into a new template. A few sentences in the existing task or research note may be enough. No separate checklist, table, report, acceptance receipt, reviewer team, or explanation of skipped steps is required by this skill.

Before a new or materially changed check

Identify the question, plausible alternative explanation, observation that distinguishes them, controls, comparison unit, authorized budget, and decision rule. Reuse an existing specification when these are already clear. Fix the rule before examining results; label retrospective or exploratory analyses honestly.

For empirical work, check the metric direction, independent units, and evaluation protocol. For theory, state assumptions and the proposition or counterexample needed. Do not require numerical metrics for a proof. Ask the owner only when missing information changes validity, authority, or material cost.

Current project instructions, ownership, resource limits, and publication rules remain authoritative. Historical reports can contain expired restrictions. This skill grants no new permissions or executor, and does not justify a run, a repeated check, or an additional agent just to demonstrate compliance.

When new evidence changes the answer

Inspect the relevant artifact. Keep execution state separate from scientific interpretation: a completed run can be negative, inconclusive, or invalid. A failed run is still a real outcome.

Verify the sources being added or reused for this conclusion, not the entire project on every turn. For empirical evidence, follow its own experiment ID and match the project, run, data, model, seed, endpoint and relevant configuration. Check host, microbatch and accumulation when they distinguish the runs. Equal titles or effective batch sizes do not establish the same experiment. Mark unknowns and look for newer attempts before treating an old limitation as current.

For a derivation, check the proposition, assumptions, argument and gaps. For literature, follow the paper and source location; paraphrases are valid. Operational observations use their own record type. These do not need invented experiment metadata.

State what the result supports and what remains open. Failure to establish equivalence does not by itself establish non-equivalence. If the interval crosses an equivalence boundary, do not turn a failed confirmation gate into a refuted population claim. Distinguish failure of a registered decision rule from proof of the opposite claim. Likewise, non-significance and overlapping intervals do not establish equivalence. Preserve scope and uncertainty instead of forcing a terminal scientific status.

Save new reusable findings through the project's existing knowledge contract, with actual state, sources, limitations and relevant links. Reuse correct existing records. Read new or changed records back, including their relationships. A returned ID or a completed card does not prove a claim. If publication is unavailable, keep one pending item in the existing task or note and continue independent work; follow the normal retry path rather than starting a loop.

When several results affect the next decision

Compare only relevant results and competing explanations. Separate changed conditions, reused data, invalid attempts and independent replications. Use a table only if it clarifies the comparison. Keep earlier results as history.

Explain what changes in the interpretation. Choose another test only if needed: which predictions it distinguishes, why existing artifacts cannot answer it, and whether a cheaper analysis suffices. Narrowing a claim or stopping a branch can be the right outcome. Statistical analysis or a full report belongs in the existing specialist workflow when the task needs it, not after every small update.

Optional research map

A research map is a dated view, not an additional ledger that must mirror every result. Use it when explicitly requested or when the current question and branches benefit from it. Otherwise keep the reasoning in the existing task, research note and shared records. No map means no map-maintenance debt. Do not invent causal nodes to fit a tree.

When updating a map, reuse stable IDs and source links, set certainty explicitly, and use versioned narrow updates. Preserve other writers' questions and substantive history without quotas. If safe update is unavailable, leave that requested update pending and continue research. Do not bypass the writer or block productive work for presentation maintenance.

Finish

Update the existing task or canonical note with the new finding and next justified action when there is one. Do not duplicate unchanged content across stores or add routine records solely because a skill was loaded. No adoption score or time log is required per turn. Evaluate this protocol retrospectively through actual changed decisions, not numbers of documents or skill reads.

Origin

Original integration informed by an audit of ResearcherOS, especially experiment planning, result interpretation and cross-experiment synthesis. No upstream skill text or implementation is bundled.